Addressing the challenge of precise customer value segmentation in multilingual customer service systems for business English, this paper primarily analyzes the application of an enhanced RFM model and optimized clustering algorithms within such systems. The study first constructs an RSA customer value assessment model, using the Recency (R), Stability (S), and Average Spending (A) as core indicators. The CRITIC weighting method is employed to scientifically determine the weight of each dimension, overcoming the issues of indicator collinearity and uniform weighting in traditional RFM models. Building upon this foundation, an improved K-means algorithm based on K-nearest neighbors and density peaks was designed. By optimizing the selection of initial cluster centers, the algorithm’s convergence speed and classification accuracy were enhanced. Empirical validation using 366 active customers from a company identified four distinct customer segments: 43 high-value loyal customers, 127 high-potential new customers, 142 latent-value customers, and 54 low-value general customers. Analysis of variance revealed highly significant differences across all clusters in R, S, and A metrics, with F-values of 80.014, 92.816, and 117.607 respectively (p<0.001), confirming the statistical validity of the segmentation results. System performance testing demonstrated excellent efficiency and stability, with a response time of only 34.27 seconds even when processing 10,000 records.
